Purification of endohedral and other fullerenes by chemical methods
Patent 7208132 Issued on April 24, 2007. Estimated Expiration Date: 
October 1, 2022. Estimated Expiration Date is calculated based on simple USPTO term provisions. It does not account for terminal disclaimers, term adjustments, failure to pay maintenance fees, or other factors which might affect the term of a patent.
October 1, 2022. Estimated Expiration Date is calculated based on simple USPTO term provisions. It does not account for terminal disclaimers, term adjustments, failure to pay maintenance fees, or other factors which might affect the term of a patent.Patent ReferencesIsolation of small-bandgap fullerenes and endohedral metallofullerenes Isolation of small-bandgap fullerenes and endohedral metallofullerenes Patent #: 6517799 InventorsAssigneeApplicationNo. 10263374 filed on 10/01/2002US Classes:423/461, Recovery or purification423/445B, Fullerene (e.g., C60, C70, etc.)977/845, Purification or separation of fullerenes or nanotubes205/687, ELECTROLYTIC MATERIAL TREATMENT (PRODUCT, PROCESS, AND ELECTROLYTE COMPOSITION)558/87ProcessesExaminersPrimary: Hendrickson, Stuart L.Attorney, Agent or FirmInternational ClassC01B 31/00DescriptionBACKGROUND OF THE INVENTION This invention relates to purification of fullerenes, especially endohedral fullerenes, by chemical methods. At the present time, endohedral fullerenes are available in much smaller quantities than the empty fullerenes. The major production method used to generate macroscopic quantities of empty fullerenes and endohedral fullerenes is the carbon-arcdischarge method first revealed by Kratschmer and Huffman (Kratschmer, 1990). This method produces a carbon soot containing approximately 10 20% fullerene materials, the majority of which is C60 and C70. Of this 10 20%, only ca. 1% of thefullerenes are of the metallo-endohedral variety. Because the empty and endohedral fullerenes are formed together as an intimate mixture, it is necessary to first separate them from each other before investigations of their individual properties canproceed. Chemical aspects of endohedral metallofullerenes and their purification have been the subjects of several published review articles (Bethune, 1993; Nagase, 1996; Liu, 2000, Shinohara (2000)). Many scientific papers and reports concern thepurification and separation of endohedral metallofullerenes from empty fullerenes. The great majority of these papers and reports involve chromatographic techniques, particularly high performance liquid chromatography (HPLC). Three of the key papers describing the first HPLC purifications of endohedral metallofullerenes areShinohara et al. (Shinohara, 1993) for Sc2@C74, Sc2@C82, and Sc2@C84, Kikuchi et al. (Kikuchi, 1993) for La@C82, and Yamamoto et al. (Yamamoto, 1994) for La@C82. The most productive HPLC protocols now usedual-stage chromatographic procedures; a recent report detailing this is Okazaki et al. (Okazaki, 2000) highlighting the HPLC separation of Sm@C2n species. This dual-stage HPLC procedure uses two commercially available HPLC columns in series, the"Buckyprep" (Nakalai Tesque, Japan) and the "Buckyclutcher" (Regis Chemical, USA) columns, both of which are special columns developed for the separation of fullerene compounds. Meyerhoff (Xiao, 1994) reported that a derivatizedtriphenylporphyrin-silica HPLC stationary phase was efficient for the one-stage separation of La@C82 and Y@C82 from empty fullerenes. Various solvent extraction procedures have been devised by which the content of endohedral fullerenes are enriched relative to the empty fullerenes. Typically these procedures use polar solvents to extract higher ratios of endohedral fullerenesrelative to the empty fullerenes in as-produced fullerene soot. Examples of solvents used and endohedral fullerenes extracted include: N-N-dimethylformamide for M@C82 (M=Ce, Ding, 1996; M=Gd, Sun, 1999) Aniline for M@C60 (M=Y, Ba, La, Ce, Pr,Nd, Gd, Kubozono, 1996) Pyridine for mixed M@C2n (M=La, Ce, Liu, 1998). Similar solvent extractions have been conducted prior to HPLC separations of endohedral metallofullerenes to increase the efficiency of the chromatographic operation. Tso etal. (Tso, 1996) reported a non-chromatographic procedure for enrichment of mixed Scm@C2n. A solid phase extraction with C18-bonded silica was performed on a chlorobenzene solution of empty fullerenes and Scm@C2n; 20 to 30% ofthe empty fullerenes were selectively removed by this technique. Sublimation techniques that partially enrich the endohedral fullerene content of mixed fullerene materials have been reported. Yeretzian et al. (Yeretzian, 1993) reported a gradient sublimation method for partial enrichment of La@C82. Diener et al. (Diener, 1997) conducted a variable temperature sublimation study of La@C2n and U@C2n. Cagle et al. (Cagle, 1999) reported a dual-temperature sublimation enrichment of HOm@C82. Ogawa et al. (Ogawa, 2000) recentlyreported a vacuum-sublimation enrichment of Er@C60. The preparation of discrete ionic fullerene compounds in condensed media has been extensively reviewed (Reed, 2000(a)). This review covers the relevant literature from the discovery of the macroscopic production of fullerenes (ca. 1990) to thepresent. Synthesis and full experimental details for the isolation of salts of the fullerene cations C76.sup. (Bolskar, 1996; Bolskar, 1997), C70.sup. (Bolskar, 1997), C60.sup. (Bolskar, 1997; Reed, 2000(b)) and HC60.sup. (Reed2000(b)) were reported as indicated. Tumanskii et al. (Tumanskii, 1998) reported the in-situ detection of proposed H(La@C82.sup. ) and H(Y@C82.sup. ) species. U.S. Pat. No. 6,303,016 to Diener et al. discloses a method for extracting small bandgap fullerenes from a mixture based on reducing the small bandgap fullerenes until they can be solvated. Once dissolved, the small band gap fullerenes can berecovered by returning them to their charge neutral state. In a similar method, the electrochemical separation of the "small-bandgap" fullerene C74 from other empty "large-bandgap" fullerenes by a reductive electrochemical procedure was performed(Diener, 1998). In that procedure, the anion C74- was selectively produced as it reduces at a more positive potential than most of the empty fullerenes, due to its unusually high electron affinity. The C74- anion was then separatedfrom the neutral fullerenes and finally plated out as a neutral species on a platinum electrode by bulk electrochemical single-electron oxidation. This procedure was also demonstrated in conjunction with endohedral Gd-metallofullerenes, producing amaterial enriched in Gd@C2n and C74 but deficient in large-bandgap empty fullerenes. The basis for this procedure is the notably higher first reduction potential of C74 (and Gd@C2n) as compared to the empty fullerenes. In principle,a similar reductive protocol could be conducted to separate endohedral metallofullerenes from empty larger-bandgap fullerenes or to separate fullerenes larger than C70 from the highly abundant C60 and C70. Some time after the isolation of discrete fulleride (reduced fullerene) species, reversible electrochemical detection of empty fullerene cations (or fullereniums), i.e., fullerenes with one electron removed from the HOMO level of the neutralmolecule, was achieved. This required more rigorous conditions and different solvents than those needed for the electroreduction experiments. Because the HOMO levels of common, empty fullerenes are relatively low-lying in energy, the fullerene cationspecies formed by removal of an electron from this HOMO are highly electrophilic. This property makes the fullereniums highly reactive toward nucleophiles. For these reasons, fullerene cations are very reactive toward even trace quantities of solventimpurities like water, so very high solvent purity and strictly water free conditions are preferred for their production. Preferred solvents are non-nucleophilic and weakly coordinating. Chlorinated alkane and arene solvents were found to be suitablynon-reactive towards the transient cations produced in the electrochemical experiments. One-electron oxidation potentials for the empty fullerenes are summarized in the E1Ox column of Table 1. Note that the most abundant fullerenes C60and C70 are more difficult to oxidize than the scarcer higher fullerenes. The high oxidation potentials for oxidation (greater than ca. 0.7 V vs. Fc/Fc.sup. ) confirm that these fullerene cations are highly reactive species. Since the timescale of electrochemical generation and detection of fullerenium ions is on the order of seconds, these results did not guarantee that fullerene cations could be prepared and isolated as discrete salts. Fullerenium cation synthesisis a greater synthetic challenge than the isolation of discrete fulleride anions. Bolskar et al. (Bolskar, 1996) state "the problem of oxidizing fullerenes to fullerene cations is the problem of finding an oxidant strong enough for the task, but onewhich does not bring along with it a reactive nucleophile." The oxidant must have enough thermodynamic driving force to abstract an electron from the neutral fullerene. Once accomplished, the now reduced oxidant must not subsequently react with thenewly formed fullerene cation. The oxidant reagent also needs to deliver a counter-anion to balance the newly formed positive charge of the fullerene cation. This anion also must be inert and not react with the fullerene cation in any manner neither bynucleophilic attack and addition nor by electro-reducing the fullerene. The first synthesis of a fullerenium cation salt, that of [C76.sup. ][CB11H.sub.6Br.sub.6-] (Bolskar, 1996), is illustrative of these principles. C76 was chosenas the first test case for fullerene oxidation because its oxidation potential is lower than that of C60 (and C70) and it is commercially available as the pure D2 isomer. The synthesis of [C76.sup. ][CB11H.sub.6Br.sub.6-]proceeded in the following manner (Equation 1): C76 [(2,4-Br2C.sub.6H.sub.3)3N.sup. ][CB11H.sub.6Br.s- ub.6-]→[C76.sup. ][CB11H.sub.6Br.sub.6-] (2,4-B- r2C.sub.6H.sub.3)3N (1) The oxidizing agent in this case was the radical cation of tris(2,4-dibromophenyl)amine. With an E° of 1.16 V, it has sufficient oxidizing power to remove an electron from C76 (E1Ox= 0.81 V). The neutral amine reductionby-product is very weakly basic for steric and electronic reasons and does not react with the newly formed fullerene cation. The hexabrominated "carborane" [CB11H.sub.6Br.sub.6-] anion of the oxidizing reagent, which becomes the counterion toC76.sup. , is perhaps one of the least nucleophilic and least reactive anions known (Reed, 1998) and does not react with the fullerenium ion. The hexahalogenated carborane anions ([CB11H.sub.6X.sub.6-], X=F, Cl, Br, I) have been shown tobe exceptionally weakly nucleophilic and weakly coordinating such that a variety of extremely electrophilic cations once thought impossible to isolate can now be synthesized. An excellent example of this is the recent synthesis (Reed, 2000(b)) of thefullerenium radical cations C60.sup. and C70.sup. via a very strong oxidant paired with the hexachlorinated carborane counter-anion, [CB11H.sub.6Cl.sub.6-]. The new solid superacid [H.sup. ][CB11H.sub.6Cl.sub.6-]incorporating this exceptionally inert anion was also synthesized and used to protonate C60 (Reed, 2000(b)). The stable and fully characterized compound [HC60.sup. ][CB11H.sub.6Cl.sub.6-] is the first example of an isolatedprotonated fullerene species. These accomplishments of extreme oxidation and protonation, i.e., the synthesis of C60.sup. and HC60.sup. in condensed media, serve to demonstrate that fullerenes can in fact be protonated and oxidized and thatthe resulting cationic fullerene species are stable and can be handled and isolated under appropriate experimental conditions. There remains a need in the art for improved methods for purifying fullerenes, especially endohedral fullerenes. The present invention provides a new, alternative purification strategy that exploits strong chemical property differences betweenthe empty and endohedral. SUMMARY OF THE INVENTION Fullerenes are typically obtained from soot obtained by combustion or electric-arc methods. C60 is the most abundant fullerene in the soot. Such soots also can contain minor amounts of other fullerenes, such as, higher fullerenes,metallofullerenes, endohedral fullerenes and so on. Purifying minor components from complex mixtures is the stereotypical recalcitrant problem faced by many scientists. In this invention, the problem of purifying minor components was solved bychemically modifying the desired fullerenes. The chemical modification of the desired fullerenes provides a chemical property difference between the desired and undesired fullerenes which allows for separating the fullerenes. For example, the inventionallows purification and separation of endohedral fullerenes from empty fullerenes and thereby resolves a long-standing problem in fullerene science. The invention provides a method of separating a first and a second set of fullerenes in a fullerene mixture comprising the steps of providing a fullerene mixture comprising a first set and a second set of fullerenes; forming a stable fullerenecation of one of the first or the second set of fullerenes, wherein the fullerene cation is formed in a solvent; and separating the fullerene cation from the other set of fullerenes. The selective formation of cation species is performed by chemicaloxidation, electrochemical oxidation, or chemical addition of cationic electrophilic groups and is followed by a recrystallization or precipitation procedure by which the cationic species are separated from the neutral species. The method of the invention is applicable and adaptable to the separation and purification of different classes of fullerenes, including but not limited to metal and non-metal containing endohedral fullerenes, higher empty fullerenes, exohedrallyderivatized empty and endohedral fullerenes, empty and endohedral heterofullerenes, and combinations thereof. In an embodiment, the invention relates to a method of purifying fullerenes based on differences between the oxidation potentials of specific classes of fullerenes or specific fullerenes. In particular, arc synthesis of endohedral fullerenesproduces a mixture of fullerene material. The fullerene material is a complex mixture of a variety of specific fullerenes, of which C60 and C70 are the most abundant. The C60 and C70 fullerenes have an oxidation potential thatdiffers from many of the minor constituents of the fullerene material such as endohedral fullerenes and higher fullerenes. Oxidation of fullerene material in a solvent with an appropriate oxidizing agent selectively oxidizes endohedral fullerenes,higher fullerenes, and mixtures thereof, while leaving C60 and C70 neutral. The oxidized fullerenes are then precipitated or recrystallized by decreasing the polarity of the solvent, while the C60 and C70 fullerenes remain soluble. The solvent is separated from the precipitated oxidized fullerene and the oxidized fullerene can then be reduced back to the unoxidized fullerene with an appropriate reducing agent. In an embodiment, the invention relates to a method of purifying selected fullerenes from a fullerene material. The method comprises the steps of: (a) providing a fullerene material, the fullerene material comprising the selected fullerenes andundesired fullerenes; (b) forming a stable fullerene cation of one of the selected fullerenes or the undesired fullerenes in a solvent; (c) decreasing the polarity of the solvent to precipitate the fullerene cation; and (d) separating the solvent fromthe precipitated fullerenes. In one embodiment, the selected fullerenes have an oxidation potential that differs from undesired fullerenes of the fullerene material. In a preferred aspect of this embodiment, the selected fullerenes have an oxidation potential that differsfrom the oxidation potential of the undesired fullerenes of the fullerene material by 0.10 V (using the ferrocene standard). In another preferred aspect of this embodiment, the selected fullerenes have an oxidation potential that differs from undesiredcomponents of the fullerene material by 0.30 V using the ferrocene standard. In a preferred aspect of this embodiment, the selected fullerenes are endofullerenes, higher fullerenes, metallofullerenes or mixtures thereof. In another embodiment, the invention relates to a method of enhancing the purity of selected fullerenes. The method comprises the steps of (a) providing a fullerene material, the fullerene material comprising the selected fullerenes andundesired fullerenes; (b) forming a stable fullerene cation of one of the selected fullerenes or the undesired fullerenes in a solvent; (c) decreasing the polarity of the solvent to precipitate the fullerene cation; and (d) separating the solvent fromthe precipitated fullerenes, thereby enhancing the purity of the selected fullerenes. The selective formation of cation species is performed by chemical oxidation, electrochemical oxidation, or chemical addition of cationic electrophilic groups and is followed by a recrystallization or precipitation procedure by which the cationicspecies are separated from the neutral species. In yet another embodiment, the invention relates to a method of preparing an endohedral fullerene substantially free of C60 and C70 fullerenes comprising the steps of providing a fullerene material, the fullerene material containingC60 and C70 fullerenes and an endohedral fullerene; forming a stable fullerene cation of the endohedral fullerene, wherein the fullerene cation is formed in a solvent; decreasing the polarity of the solvent to precipitate the fullerene cation;and separating the solvent containing the C60 and C70 fullerenes from the precipitated endohedral fullerene. In yet another embodiment, the invention relates to a method of separating C60 and C70 fullerenes from fullerene material comprising the steps of providing a fullerene material, the fullerene material containing a first set of C60and C70 fullerenes and a second set of other fullerenes; forming a stable fullerene cation of the second set of fullerenes, wherein the stable fullerene cation is formed in a solvent; decreasing the polarity of the solvent to precipitate thefullerene cation; and separating the solvent containing the C60 and C70 fullerenes from precipitated second set of fullerenes The method of chemically separating minority fullerene materials from mixtures with other majority fullerene materials by exploiting particular chemical and redox properties can be used for the purification of a broad range of endohedral andempty fullerene materials. A general procedure is described herein for the separation of metal-containing endohedral fullerenes from the empty fullerenes which is applicable to fullerene materials possessing appropriate chemical and/or redoxcharacteristics including, but not limited to, the following examples: (A) Separation of M@C82 endohedral metallofullerenes from empty fullerenes and other Mm@C2n endohedral fullerenes by selective chemical or electrochemical one-electron oxidation of M@C82 species to their molecular cations,M@C82.sup. . This includes all different possible structural isomers and/or enantiomers of M@C82. (M is an element from the periodic table; the "@" symbol refers to the endohedral or interior nature of the M atom inside of the fullerene cage;m is the integer 1, 2, or 3; n is an integer number) (B) Separation of Mm@C2n endohedral metallofullerenes as individual molecular species and/or as a general class from empty fullerenes by selective chemical or electrochemical one-electron oxidation of Mm@C2n species to theirmolecular cations Mm@C2n.sup. . This includes the extraction and separation of Mm@C2n endohedrals which are highly insoluble in their neutral states but are soluble and extractable as molecular cations; many of these species havenever before been isolated. This also includes all different structural isomers and/or enantiomers of Mm@C2n. (C) Separation of specifically those insoluble M@C2n species abundant in sublimates such as M@C60, M@C70, and M@C74 with M=trivalent endohedral lanthanides (Ce, Pr, Nd, Gd, Tb, Dy, Ho, Lu and possibly Er) by selective chemicalor electrochemical one-electron oxidation of M@C2n species to their molecular cations M@C2n.sup. . In particular, the separation can be accomplished by solvent washing removal of soluble C2n and M@C2n and oxidative solubilization andremoval of oxidizable C2n and M@C2n species, leaving behind the insoluble M@C2n materials. (D) Separation of specifically those insoluble M@C2n species abundant in sublimates such as M@C60, M@C70, and M@C74 with M=divalent endohedral lanthanides (Sm, Eu, Tm, Yb) by selective chemical or electrochemical one-electronoxidation of M@C2n species to their molecular cations M@C2n.sup. . (E) Separation of empty higher fullerenes C2n (those larger than C60 and C70) as a general class from C60 and C70 by selective chemical or electrochemical one-electron oxidation of C2n species to their molecularcations, C2n.sup. (in the absence of endohedral fullerenes). This includes all different possible structural isomers and/or enantiomers of C2n. (F) Separation of empty fullerenes and endohedral fullerenes exohedrally derivatized by functional groups that are selectively oxidizable by chemical or electrochemical means to form cationic derivatized fullerene and endohedral fullerenespecies. The resulting oxidized component is the fullerene cage structure, the exohedral functionalization, or both components together. (G) Separation of skeletally substituted empty and endohedral fullerenes (e.g. heterofullerenes) from non-substituted fullerenes and endohedral fullerenes by selective chemical or electrochemical oxidation of those heterofullerenes which have oneor more individual carbon atom(s) of their fullerene shells replaced by elements other than carbon, to their molecular cations. (H) Separation of non-metal containing endohedral fullerenes (e.g. Nm@C2n, Pm@C2n, etc.) from empty fullerenes by selective chemical or electrochemical one-electron oxidation of the given endohedral species to itscorresponding molecular cation. This invention is also applicable to the separation of mixed multi-element containing endohedral fullerenes, e.g., Sc3N@C80, ErSc2N@C80, etc. (I) Achievement of all the above categories of separation (A through H) based on selective formation of cationic fullerene species by chemical protonation or addition of other cationic electrophilic groups, which is distinct from fullerene cationformation via the aforementioned chemical or electrochemical oxidation. Cation formation can equally be conducted by oxidative electrochemistry or by chemical addition of a cationic agent, such as protonation by a Brφnsted acid or addition of anelectrophile. Photochemical cation generation methods can be used in the invention and are considered a subset of general chemical methods to oxidation and protonation. The instant invention describes general methods for the production of new cationic fullerene species. These fullerene cations can subsequently be put to other uses, for example, they can be used in ensuing derivitization reactions withnucleophiles to produce exohedrally derivatized fullerene and/or endohedral fullerene compounds that may not be obtained through other routes. The chemical purification methods described herein are also be applicable to fullerene materials produced by new production methods other than the standard arc-discharge and laser ablation methods of producing fullerenes. For example, thedescribed chemical purification procedures can be applied to any new production method that produces materials containing higher yields of endohedral fullerenes than currently available. The general methods described herein will allow the large-scaleisolation of pure endohedral metallofullerene compounds in a manner not possible prior to this disclosure. The invention presents a significant improvement over prior separation and purification technologies, the chief of which is high-performance liquid chromatography (HPLC). The invention exploits strong chemical property differences between thefullerene materials to be separated. As HPLC separations of fullerene materials typically depend on chemical properties that are similar amongst fullerenes, they therefore require substantial time and economic input. The invention also uses commonlyavailable equipment, unlike HPLC, and commonly available solvents and reagents. The disclosed invention is also more amenable to a bulk scale-up compared to chromatographic techniques. The disclosed procedures can also be used as preparatory stepsprior to chromatographic techniques, greatly increasing their efficiency. Advantageously the method of the invention produces the final purified product(s) in the solid state, thus avoiding difficult removal of low vapor pressure and high-boilingsolvents. The disclosed oxidative extraction procedure is a significant improvement over the prior technology in that it allows efficient extraction of Mm@C2n which are normally insoluble and unextractable as neutral species (or soluble orextractable at best with a very low efficiency) as cations. This increases the total available yield of endohedral fullerenes. This is an improvement over the various solvent extraction procedures that use highly polar and/or coordinating solvents asagents of partial preferential extraction. Such solvents extract endohedral fullerenes at lower efficiencies than the disclosed invention and as mixtures with empty fullerenes. Also, this invention avoids the use of these highly polar and/orcoordinating solvents, which may irreversibly bind in an unknown manner exohedrally to endohedral metallofullerenes. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 shows a comparison of the frontier orbital levels of M@C82.sup. , M@C82, and M@C82- (where M is in its 3 state, donating 3 electrons as in the formalism M3 @C823-). FIG. 2 shows a LD-TOF MS of mixed Gd-endohedral fullerene sublimate. FIG. 3 shows a LD-TOF MS of Gd@C82 separated according to Scheme 1. FIG. 4 shows a Vis-NIR electronic absorption spectrum of Gd@C82.sup. in ODCB solution. FIG. 5 shows a differential pulse voltammogram of Gd@C82.sup. in ODCB solution. FIG. 6 shows a LD-TOF MS of Gdm@C2n separated according to Scheme 2. FIG. 7 shows a differential pulse voltammograom of Gdm@C2n in ODCB solution. FIG. 8 shows the LD-TOF MS of Ym@C2n.sup. separated according to Scheme 2. FIG. 9 shows a differential pulse voltammogram Of Ym@C2n.sup. in ODCB solution. FIG. 10 shows a LD-TOF MS of insoluble Gd@C2n separated according to Scheme 3. FIG. 11 shows a LD TOF-MS of ODCB-insoluble Gd@C60 class. FIG. 12 shows a LD TOF-MS of the AlCl3-treated ODCB-insoluble Gd@C60 class. FIG. 13 shows a LD-TOF MS of mixed Tm-endohedral fullerene sublimate. FIG. 14 shows a LD-TOF MS of purified Tmm@C2n.sup. separated as soluble cations from insoluble endohedrals and empty C2n fullerenes. FIG. 15 shows a LD-TOF mass spectrum of the soluble oxidizable Tm@C60 class. DETAILED DESCRIPTION OF THE INVENTION Unless otherwise stated, the following terms are accorded the following definitions. As used herein the term "fullerene material" generally refers to a material containing one or more specific fullerenes. Typically the term "fullerene material" refers to a raw or crude preparation of fullerenes, i.e., soot, although this termcan also be used in reference to raw or crude preparations of fullerenes that have been further processed yet contain a mixture of specific fullerenes. As used herein the term "fullerene" can refer to a specific fullerene or class of fullerenes that can be distinguished from other fullerenes based on a physical or chemical property, i.e., oxidation potential or reduction potential. If thefullerene is a class of fullerenes rather than a specific fullerene, the physical or chemical properties may have a range of values, i.e., the oxidation potential of the set may cover a range of values. In a separation process, one fullerene in afullerene material may be termed a "selected" or "desired" fullerene while the other fullerene(s) may be termed "rejected" or "undesired" fullerene(s). Fullerenes suitable for use with the present invention include, but are not limited to, emptyfullerenes, derivatized empty fullerenes, empty higher fullerenes, empty small-bandgap fullerenes, heterofullerenes, endohedral fullerenes, and derivatized endohedral fullerenes, and giant fullerenes where N is 50 or more which may be nested and/ormulti-concentric fullerenes As used herein, the term "cation" may refer to one or more cations. As used herein the term "stable cation" refers to a cation which does not revert to a neutral state or irreversibly react under the conditions of its formation. As used herein the term "oxidized fullerene" refers to fullerenes to which an electron was removed from or to fullerenes to which a positive charge was added. As used herein an "easily oxidizable fullerene" is a fullerene with an oxidationpotential of less than or equal to about 0.8V (vs. the ferrocene/ferrocenium couple). As used herein the term "endohedral fullerene substantially free of C60 and C70 fullerenes" refers to a preparation of endohedral fullerene which is less than 20 percent by weight of C60 and C70 fullerenes. Preferably,"endohedral fullerene substantially free of C60 and C70 fullerenes" refers to a preparation of endohedral fullerene which is less than 20 percent by weight of C60 and C70 fullerenes. In general, fullerenes have "low solubility" in all solvents compared to most organic compounds. For this application, we define relative solubility with the terms soluble, moderately soluble, slightly soluble, substantially insoluble andinsoluble on a scale based on the known solubility of fullerene C60 in various solvents (See: Ruoff et al., 1993). C60 is insoluble in polar solvents like acetonitrile and methanol (no measurable amount dissolves). C60 is substantiallyinsoluble in the polar solvent tetrahydrofuran (~0.001 mg/mL or less). C60 is slightly soluble in alkane hydrocarbons (0.002 to 2 mg/mL) and haloalkanes (0.1 to 5 mg/mL). C60 is slightly soluble to soluble in various arenes (rangingfrom 1.7 mg/mL in benzene to 27 mg/mL in 1,2-dichlorobenzene). C60 is soluble in substituted naphthalenes (30 to 50 mg/mL). The definitions used herein are thus, insoluble approximately 0 mg/mL; substantially insoluble greater than 0 mg/ml andless than or equal to about 0.001 mg/mL (this represents trace amounts); slightly soluble between about 0.002 to about 5 mg/mL; moderately soluble greater than 5 mg/mL and less than about 20 mg/mL; and soluble greater than about 20 mg/mL. As notedabove, insoluble classes of fullerenic species are also insoluble in non-reactive polar solvents. Those classes of fullerenes that are soluble in common solvents (e.g., C60) are either insoluble, substantially insoluble or slightly soluble in polarsolvents depending on the fullerene in question. Fullerenes and endohedral fullerenes (those fullerenes encapsulating an atom or atoms in their interior space) are a recently discovered class of chemical compounds. Fullerenes are all-carbon molecular cages with no dangling or terminal bonds. C60 is the most abundant fullerene, with C70 the next most common. Larger C2n species, i.e., C76, C78, C82, C84, etc., (2n between 70 and 100) are called the "higher fullerenes." A distinct class of empty fullerenemolecules are those with small HOMO-LUMO gaps such as C74 (D3h) and C80(Ih), which behave as small-bandgap materials in the solid state (Diener, 1998). These empty fullerenes may be termed "small band-gap fullerenes". The term "fullerene" is used generally herein to refer to any closed cage carbon compound containing both six-and five-member carbon rings independent of size and is intended to include the abundant lower molecular weight C60 and C70fullerenes, larger known fullerenes including C76, C78, C84 and higher molecular weight fullerenes C2N where N is 50 or more (giant fullerenes) which may be nested and/or multi-concentric fullerenes. The term is intended to include"solvent extractable fullerenes" as that term is understood in the art (generally including the lower molecular weight fullerenes that are soluble in toluene or xylene) and to include higher molecular weight fullerenes that cannot be extracted, includinggiant fullerenes which can be at least as large as C400. Additional classes of fullerenes include, among others specifically noted herein, endohedral fullerenes contains one or more elements, particularly one or more metal elements, andheterofullerenes in which one or more carbons of the fullerene cage are substituted with a non-carbon element, such as B or N. The term fullerenic material is used generally to refer to a material that contains a mixture of fullerenes or a mixture of oneor more fullerenes with non-fullerenes, e.g., amorphous carbonaceous materials that may for example be formed during fullerene synthesis by any known method and includes raw or crude preparations of fullerenes, such as combustion soot as well as raw orcrude preparations of fullerenes that have been at least partially purified, for example, by extraction and/or sublimation. Fullerenes are members of a broader class of materials called "carbon nanomaterials" which as used herein generally refers to any substantially carbon material containing six-membered rings that exhibits curving of the graphite planes, generallyby including five-membered rings amongst the hexagons formed by the positions of the carbon atoms, and has at least one dimension on the order of nanometers. Examples of carbon nanomaterials include, but are not limited to, fullerenes, single-walledcarbon nanotubes (SWNTs), multiple-walled carbon nanotubes (MWNTs), nanotubules, and nested carbon structures with dimensions on the order of nanometers. Carbon nanomaterials may be produced in soot and, in certain cases, carbon nanomaterials may beisolated from the soot or enriched in the soot. Soot produced during the synthesis of carbon nanomaterials, such as fullerenes, typically contains a mixture of carbon nanomaterials which is a source for further purification or enrichment of carbonnanomaterials or which may itself exhibit desired properties of carbon nanomaterials and be useful as an addition to convey those properties. The term "carbon nanomaterials," when used without limitation, is intended to include soot containingdetectable amounts of carbon nanomaterials. For example, the term fullerenic soot is used in the art to refer to soot containing fullerenes. Fullerenic soot is encompassed by the term carbon nanomaterials. Non-fullerenic carbonaceous materialsinclude, but are not limited to, non-fullerenic carbon nanomaterials as well as amorphous carbonaceous materials. Carbon nanomaterials are not amorphous carbonaceous materials. Endohedral fullerenes are fullerene cages that encapsulate an atom or atoms in their interior space. They are written with the general formula Mm@C2n, where M is an element, m is the integer 1, 2, 3 or higher, and n is an integernumber. The "@" symbol refers to the endohedral or interior nature of the M atom inside of the fullerene cage. Endohedral fullerenes corresponding to most of the empty fullerene cages have been produced and detected under varied conditions. Endohedralmetallofullerenes useful for the present invention, include, but are not limited to those where the element M is a lanthanide metal, a transition metal, an alkali metal, an alkaline earth metal, and a radioactive metal. Other endohedral fullerenesuseful for the present invention are those containing radioactive elements. Radioactive elements include naturally occurring and non-naturally occurring isotopes. Radioactive elements also include nonmetals and semimetals such as iodine, phosphorus,bismuth, antimony, arsenic and tin as well as metals such as the actinide metals. Endohedrals containing lanthanide metals have by far received the most attention to date, although studies on non-metallo endohedrals such as He@C60, N@C60,P@C60, etc. are developing. Of the Mm@C2n metallo-endohedrals, the ones having the particular formulation M@C82 are the subject of most investigations to date. The study of other M@C2n and multiple-metal containingMm@C2n materials is proceeding at a slower rate because of the very low availabilities of the isolated materials needed for characterization. The invention provides a method of separating a first and a second fullerene in a fullerene material comprising the steps of: (a) introducing the fullerene material into a solvent; (b) forming a stable fullerene cation from the first fullerene,wherein the solvent is selected so that at least one of the fullerene cation or the second fullerene is soluble therein; and (c) separating the fullerene cation or a salt thereof from the second fullerene. Optionally, the method may include the step of returning the cation to a neutral state. The selective formation of cationic fullerene species is performed by chemical oxidation, electrochemical oxidation, or chemical addition of cationic electrophilic groups. Cation formation can equally be conducted by oxidative chemistry or bychemical addition of a cationic agent, such as protonation by a Brφnsted acid or addition of an electrophile. Photochemical cation generation methods can be used in the invention and are considered a subset of general chemical methods to oxidationand protonation. The selective formation of cationic fullerene species is followed by a separation procedure. The separation procedure may be accomplished in several ways. In one embodiment, the second fullerene and the fullerene cation are both substantiallysoluble in the solvent and fullerene cation is separated from the second fullerene by decreasing the polarity of the solvent to precipitate a salt of the fullerene cation and separating the fullerene cation salt precipitate from the solvent. The firstfullerene may be soluble in the solvent or not. This procedure exploits the solubility differences of cationic fullerene compounds and neutral fullerene compounds. Ionic compounds are soluble in solvents with a higher polarity and higher dielectricconstants, whereas the opposite is generally true for neutral compounds. This can be used to separative advantage. Some solvents like o-dichlorobenzene (ODCB) are capable of dissolving fullerene anions, cations, and neutral species. When a non-polarsolvent such as the hydrocarbon hexane is added to an ODCB solution of mixed fullerene neutral and cationic molecules, the polarity is drastically reduced, causing the cations to precipitate as a solid salt while the majority of the neutrals remain insolution. In another embodiment, the separation procedure is as follows. The first and second fullerene are substantially soluble in the solvent, the fullerene cation is substantially insoluble in the solvent, and the fullerene cation is separated fromthe second fullerene by precipitating a salt of the fullerene cation and separating the fullerene cation salt precipitate from the solvent. Precipitation occurs upon the oxidative cation formation. In another embodiment, the separation procedure is as follows. The first and second fullerene are substantially insoluble in the solvent, the fullerene cation is substantially soluble in the solvent, and the fullerene cation is separated fromthe second fullerene by separating the insoluble second fullerene from the solvent. Oxidation of Fullerenes Empty fullerenes generally are easy to reduce (reduction=addition of electron(s)) while conversely being difficult to oxidize (oxidation=removal of electron(s)). The molecule C60 is the prototypical example of this. With its low-lying,triply degenerate HOMO level, C60 can accept up to six electrons. In solution six electrochemically reversible reductions were detected and their potentials measured by cyclic voltammetry and various pulsed techniques. Further electrochemicalstudies revealed that C70 and the higher fullerenes display similar behavior. Data for up to the fourth reduction level of selected empty fullerenes is summarized in Table 1 below. TABLE-US-00001 TABLE 1 Selected Redox Parameters of Empty Fullerenes in Solution vs the Ferrocene Standard (ferrocene/ferrocene.sup. = 0.0 V, cyclic voltammetry unless otherwise noted). Fullerene Solventa E1Red E2RedE3Red E.- sub.4Red E1Ox Reference C60 PhMe/ -0.98 -1.37 -1.87 -2.35 Xie 1992 MeCNb TCEd -1.06 1.26 Xie 1993 C70 PhMe/ -0.97 -1.34 -1.78 -2.21 Xie 1992 MeCNb TCEd -1.02 1.20 Xie 1993 C76 PhMe/-0.83 -1.17 -1.68 -2.10 Selegue 1994 MeCNe TCEd,f -0.83 -1.12 0.81 Yang 1995 C2v-C.sub.78 PhMe/ -0.72 -1.08 -1.79 -2.16 Selegue 1994 MeCNe TCEd,f -0.77 -1.08 0.95 Yang 1995 D3-C.sub.78 CH2Cl.sub.2 -0.64 -0.94 -1.70-2.05 Boudon 1998 TCEd,f -0.77 -1.08 0.70 Yang 1995 C2-C.sub.82 ODCBh -0.69 -1.04 -1.58 -1.94 0.72 Suzuki 1996 C84i TCEd,f -0.67 -0.96 -1.27 0.93 Yang 1995 C86j ODCBh -0.58 -0.85 -1.60 -1.96 0.73 Suzuki1996 aSupporting electrolyte 0.1 M Bu4NPF.sub.6.bAt -10° C. C5.4:1 PhMe:MeCN. dTCE = 1,1',2,2'-tetrachloroethane. e8:2 PhMe:MeCN. fOsteryoung square wave voltammetry (OSWV). gAt -15° C.hODCB = o-dichloro-benzene. iIsomer(s) not resolved. jMajor isomer. Fullerenium cation synthesis is a greater synthetic challenge than the isolation of discrete fulleride anions. The oxidant must have enough thermodynamic driving force to abstract an electron from the neutral fullerene. Once accomplished, thenow reduced oxidant must not subsequently react with the newly formed fullerene cation. The oxidant reagent also needs to deliver a counter-anion to balance the newly formed positive charge of the fullerene cation. This anion also must be inert and notreact with the fullerene cation in any manner neither by nucleophilic attack and addition nor by electro-reducing the fullerene. Typically, fullerene cations are very reactive toward even trace quantities of solvent impurities like water, so very highsolvent purity and strictly water free conditions are preferred for their production. In general, of all the empty common fullerenes, C60 is the hardest to oxidize (having the highest ionization potential) and likely the most difficult to protonate. The cation C60.sup. has been synthesized via a very strong oxidantpaired with the hexachlorinated carborane counter-anion, [CB11H.sub.6Cl.sub.6-] (Reed, 2000(b)). For the oxidation of fullerene molecules less demanding than C60, there is a variety of weaker chemical oxidizing agents that are commonlyused in synthetic chemistry (Connelly, 1996). Similarly, there exist a variety of anions less exotic than the halogenated carboranes [CB11H.sub.6X.sub.6-] that are not as weakly nucleophilic and weakly coordinating but are sufficient ascounteranions for less electrophilic species (Strauss, 1993). Protonation of fullerenes larger than C60 may be less demanding than that of C60, in analogy to chemical oxidation of C60 and the higher fullerenes, but will still requireappropriate synthetic conditions including but not limited to anaerobic conditions, purified and water free solvents, non-nucleophilic solvents, non-nucleophilic anions, and acidity sufficient for protonation of the given fullerene. The chemistry of endohedral fullerene compounds is less well developed than the corresponding chemistry of the empty fullerenes, primarily because of the relative scarcity of endohedral fullerenes--they constitute only several percent of thetotal fullerene content produced by the arc discharge method and are very difficult to separate from the much more abundant empty fullerenes. Relevant to this invention are the recently reported accounts of the transient electrochemical detection of metallofullerene cations and anions in fluid solution. The data obtained in these experiments provide a relative measure of the energiesneeded to add and remove electrons to and from the endohedral metallofullerenes. Electrochemical data on M@C82 endohedral metallofullerenes (and one M2@C80 species) obtained with various techniques are summarized in Table 2. The data in Table 2 indicate that the redox properties of metallofullerenes have both similarities to and differences from those of the empty fullerenes. The endohedral metallofullerenes undergo multiple reductions, similar to the emptyfullerenes. However, their first reductions are typically at higher potentials (more positive, i.e. easier to add an electron) than the larger empty fullerenes. The endohedral metallofullerenes also display single-electron oxidations at much lowerpotentials than do the empty fullerenes. In fact, these oxidations are at potentials >1 V lower (easier) than the oxidation potentials of the fullerenes C60 and C70. Also of note in Table 2 are the electrochemical measurements performed onthe very recently isolated minor isomers of La@C82 and Pr@C82. These minor structural isomers are seen to display one-electron reduction and oxidation characteristics similar to those of their respective major isomers. Two reports describesome of the electrochemistry of multiple-metal containing endohedrals. La2@C80 reduces slightly easier than the mono-metallofullerenes and oxidizes at a slightly higher potential (Suzuki, 1995). Electrochemical measurements onSc3@C82 in the strongly coordinating solvent pyridine were also reported (Anderson, 1997). Table 2 reveals that placing a metal atom (or atoms) inside of fullerene cages alters its redox characteristics. TABLE-US-00002 TABLE 2 Selected Redox Parameters of Selected M @ C82 Endohedral Fullerenes vs the Ferrocene Standard (Ferrocene/Ferrocene.sup. = 0.0 V).a Compound E1Red E2Red E3Red E4Red E-1Ox Reference La @ C82b -0.42 -1.37 -1.53 -2.26 0.07 Suzuki 1996 Gd @ C82b -0.39 -1.38c -2.22c 0.09 Suzuki 1996 Ce @ C82b -0.41 -1.41 -1.53 -1.79 0.08 Suzuki 1996 Y @ C82b -0.37 -1.34c-2.22 0.10 Suzuki 1996 Pr @ C82b -0.39 -1.35 -1.46 -2.21 0.07 Akasaka 2000b La @ -0.48 -1.41c -2.01 -0.07 Akasaka 2000b C82 - IId Pr @ -0.48 -1.39c -1.99 -0.07 Akasaka 2000b C82 - IId La2 @ C80 -0.31-1.71 -2.13e 0.56 Suzuki 1995 aSolvent: o-dichlorobenzene (ODCB), supporting electrolyte 0.1 M Bu4NPF.sub.6, data obtained by cyclic voltammetry and differential pulse voltammetry bMajor isomer ("I"). cTwo-electron process. dMinor isomer ("II"). eIrreversible. Following the electrochemical studies, Akasaka and co-workers reported the generation of reduced and oxidized solutions of both isomers ("A" and "B") of HPLC-purified La@C82 with controlled potential electrolysis (Akasaka, 2000(a);. Akasaka, 2001). Vis-NIR spectroscopy revealed characteristic changes in the spectra of the ions versus neutral La@C82. Interestingly, the anions La@C82- were reported to be significantly stable in air, while the corresponding cationsreportedly decomposed even under inert atmosphere. The very large difference in oxidation properties between the major component of the fullerene arc product (C60 and C70) and the minor component (endohedrals, Mm@C2n) provides the physical basis for the success of thisinvention in isolating endohedral from empty fullerenes. The reason for the relatively "easy" one-electron oxidation and reduction of the endohedral metallofullerenes can be partially explained by examining their electronic structures. FIG. 1illustrates the basic molecular orbital (MO) diagram for typical M@C82 species containing a three-electron donating (trivalent) metal like those in Table 2 (the trivalent metals in the case of M@C82 are generally understood to include Sc, Y,La, Ce, Pr, Nd, Gd, Tb, Dy, Ho, Yb, Lu, U, Np and Am). Overall these molecules are neutral, but they can be formally written as zwitterions, i.e., M3 @C823-. The MO diagram reveals that these species are radicals, containing one unpairedelectron in their singly degenerate HOMO levels. FIG. 1 also comparatively illustrates what happens when the electron in the HOMO is removed (M@C82.sup. ), or when another electron is added (M@C82-). In both cases, a closed shellmolecule possessing a filled HOMO level is formed. Thus, single electron oxidation and reduction are both favorable and this helps explain why the M@C82 species are so much easier to oxidize and reduce than normal, closed-shell fullerenes. Freeradicals are generally more reactive than closed shell molecules, so the addition or removal of one electron from radical M@C82 to form closed shell, metallofullerene 1 or -1 ions gives these species an added degree of stability. Importantly,because M@C82 metallofullerenes oxidize so much more readily than C60 and C70, the cations M@C82.sup. can be preferentially formed under certain chemical conditions, even in the presence of great excesses of C60 and C70. The electronic structures and MO diagrams of Mm@C2n species other than M@C82 are less well known at this time. It is expected that most Mm@C2n have a similar radical electronic structure and relatively easy one-electronoxidation as seen with the M@C82 species, particularly in the case of trivalent metals. With endohedral divalent metals, the oxidation potentials may be higher than those for the corresponding trivalent metallo-endohedrals, but still significantlylower than the oxidation potentials of C60 and C70. Thus, a purification strategy based on one-electron oxidations is also applicable to the different forms of Mm@C2n, under the proper conditions. The less abundant structuralisomers of Mm@C2n metallofullerenes are amenable to the purification methodology of the invention as well. Distinct from the endohedrals, there is a ca. 0.5 V difference in the oxidation potential of the empty higher fullerenes as a group(E1Ox range from 0.7 to 0.9 V) from C60 and C70 (E1Ox= 1.3 and 1.2 V). Thus, oxidation can also serve as handle for the separation of empty higher fullerenes from the much more abundant C60 and C70. There exist a wide variety of chemical oxidizing and reducing agents, weakly coordinating anions, and weakly nucleophilic solvents suitable for use in the oxidative based procedure of the present invention. Common chemical redox reagents havebeen extensively reviewed (Connelly, 1996). Specific redox potential and processability are important criteria when choosing these reagents. Ag.sup. as an oxidant in the disclosed procedure is advantageous because its reduction by-product, Ag metal,is readily removed by filtration from the solution of the soluble, newly oxidized material. Other oxidizing agents effective in this, disclosed invention are the radical cations of substituted triarylaminium species, for exampletris(4-bromophenyl)aminium (p-BrC6H.sub.4)3N.sup. and related derivatives of varying oxidative strength (Connelly, 1996; Bolskar, 1996). Further chemical oxidants include but are not limited to those discussed by Connelly and Geiger(Connelly, 1996). A variety of weakly coordinating anions also exist as alternatives to hexafluoroantimonate (SbF6-) (Strauss, 1993). These include but are not limited to triflate, sulfate, perchlorate, hexachloroantimonate, hexafluorophosphate,hexachloroarsenate, hexafluoroarsenate, tetrafluoroborate, tetraarylborates and their various fluorinated substituted derivatives, tetrachloroaluminate, trifluoroacetate, carboranes of various compositions e.g. CB11H.sub.6X.sub.6- (where X=H,F, Cl, Br, I, CH3), CB11(X)12-, (where X=F, Cl, CH3), CB9H.sub.10-, Co(C2B.sub.11H.sub.9)2-, etc. Useful weakly nucleophilic solvents include but are not limited to dichloromethane, 1,2-dichloroethane, 1,1',2,2'-tetrachloroethane, bromobenzene, chlorobenzene, 1,2-dichlorobenzene, various isomers of trichlorobenzene,α,α,α-trifluorotoluene, dichlorodifluoroethane and other freons, sulfur dioxide, nitrobenzene, etc. Low-polarity solvents suitable for use in the recrystallization procedures include but are not limited to hexane, hexanes (commercialmixture), pentane, heptane and other branched and straight chain alkanes, cyclohexane, decalins (cis and trans), benzene, toluene, xylene (all isomers), mesitylene, tetralin, carbon tetrachloride, etc. Other Methods of Cation Formation The disclosed separation protocols that are based on fullerene cation formation by oxidation can alternatively proceed with cation formation via addition of a cationic electrophile, such as a proton. Formation of empty fullerene and endohedralfullerene cations via protonation is a variant of the disclosed purification procedure that is applicable in a similar manner as the oxidation protocol. This is because endohedral fullerenes have a much lower protonation threshold than the emptyfullerenes. Reed et al. (Reed, 2000(b)) have shown that C60, perhaps the fullerene most difficult to protonate, can in fact be protonated under extreme conditions. Under far less drastic conditions using a much weaker acid, preliminarymeasurements (Tumanskii, 1998) indicate that La@C82 and Y@C82 likely can be protonated by trifluoroacetic acid. Thus, selective protonation of endohedral fullerenes intimately mixed with an excess of empty fullerenes is possible, allowing fora cationic purification procedure analogous to that described for oxidation but based on chemical protonation. Protonation can also be useful for separating the higher fullerenes as a group from the more abundant C60 and C70 because the higherfullerenes will protonate selectively under conditions milder than those needed to form HC60.sup. and HC70.sup. . Fullerene cations formed by protonation can thus be separated from the unprotonated species via recrystallization, similar tohow it is performed in the oxidative protocol. Oxidation of the neutral radical endohedrals does confer a slight advantage over protonation, however. One-electron oxidation of radical M@C82 molecules forms closed shell species, whereas afterprotonation the resulting cation species may remain as a radical. Radicals are generally more reactive than closed shell species, so in certain cases a cationic purification method based on oxidation rather than protonation may be more desirable. Protonation of endohedral metallofullerenes by a discrete acid is illustrated by Equation 2. (Mm@C2n) [H.sup. ][X-]→[H(Mm@C2n).sup. ][- X-] (2) The acid [H.sup. ][X-] in Equation 2 is a protic (Brφnsted) acid with a weakly coordinating counter-anion [X-], with X- being but not limited to triflate, sulfate, perchlorate, hexfluoroantimonate, hexachloroantimonate,hexafluorophosphate, tetrafluoroborate, tetraarylborate and its various substituted fluorinated derivatives, tetrachloroaluminate, trifluoroacetate, carboranes of various compositions e.g. CB11H.sub.6X.sub.6- (where X=H, F, Cl, Br, I,CH3), CB11(X)12-, (where X=F, Cl, CH3), CB9H.sub.10-, Co(C2B.sub.11H.sub.9)2-, etc. Besides protonation, other cationic electrophilic reagents R.sup. (where R is an organic or inorganic functionality) can be incorporated into the cation formation protocol. Such reagents are reacted with endohedral metallofullerene species toform cations such as R(Mm@C2n.sup. ). Kitagawa (Kitagawa, 1999) has generated analogous RC60.sup. compounds in situ where R is a polychloroalkane moiety. Separation Procedures In one embodiment, the second fullerene and the fullerene cation are both substantially soluble in the solvent and fullerene cation is separated from the second fullerene by decreasing the polarity of the solvent to precipitate a salt of thefullerene cation and separating the fullerene cation salt precipitate from the solvent. The purification procedure of the invention exploits the solubility differences of cationic fullerene compounds and neutral fullerene compounds. Ionic compounds aresoluble in solvents with a higher polarity and higher dielectric constants, whereas the opposite is generally true for neutral compounds. This can be used to separative advantage. Some solvents like o-dichlorobenzene (ODCB) are capable of dissolvingfullerene anions, cations, and neutral species. When a non-polar solvent such as the hydrocarbon hexane is added to an ODCB solution of mixed fullerene neutral and cationic molecules, the polarity is reduced, causing the cations to precipitate as asolid salt while the majority of the neutrals remain in solution. It is noted that the solubility of empty fullerenes and endohedral fullerenes in common organic solvents is relatively low (on the order of milligrams per milliliter; Ruoff, 1993). Useful solvents for the present invention that dissolve fullerene anions, cations, and neutral species include, but are not limited to halobenzenes (ODCB, chlorobenzene, bromobenzene, etc.) and haloalkyls (e.g. 1,1,1',1' tetrachloroethane). Low-polaritysolvents suitable for use in the recrystallization procedures include but are not limited to hexane, hexanes (commercial mixture), pentane, heptane and other branched and straight chain alkanes, cyclohexane, decalins (cis and trans), benzene, toluene,xylene (all isomers), mesitylene, tetralin, carbon tetrachloride, etc. In another embodiment, the separation procedure is as follows. The first and second fullerene are substantially soluble in the solvent, the fullerene cation is substantially insoluble in the solvent, and the fullerene cation is separated fromthe second fullerene by precipitating a salt of the fullerene cation and separating the fullerene cation salt precipitate from the solvent. Useful solvents for this embodiment are non-polar organic solvents in which both the first and second fullereneare soluble. Suitable solvents include, but are not limited to, non-polar hydrocarbon (hexane, pentane, decalin, etc.) and arene (benzene, toluene, etc.) solvents. In another embodiment, the separation procedure is as follows. The first and second fullerene are substantially insoluble in the solvent, the fullerene cation is substantially soluble in the solvent, and the fullerene cation is separated fromthe second fullerene by separating the insoluble second fullerene from the solvent. Useful solvents for this embodiment include, but are not limited to, at least moderately polar solvents (such as dichloromethane) which are capable of dissolving thefullerene cation. Recovery of Neutral Fullerenes Optionally, the fullerene cation can be returned to its neutral state following separation of the fullerene cation. A parent fullerene molecule can be recovered from an oxidized cation salt by a variety of chemically or electrochemicallyreductive methods. In any such reduction stoichiometry is controlled and a relatively weak reducing agent is utilized to minimize the chance of overreduction to the fullerene anion. A parent fullerene molecule can be recovered from a protonated fullerene cation salt by reaction with base. Equation 3 describes a non-limiting example of such a neutralization procedure for a protonated fullerene cation salt. [H(Mm@C2n).sup. ][X-] NaOH→Mm@C2n [Na.sup- . ][X-] H2O (3) In the formula Mm@C2n, M includes, but is not limited to the elements listed in Table 2. It is contemplated that the described processes will work for separating M@C2n where M is a three-electron donating metal such that theendohedral is formally M3 @C2n3-. These species have the electronic structures outlined by the MO diagrams in FIG. 1. FIG. 1 is a comparison of the frontier orbital levels of M@C82.sup. , M@C82, and M@C82- (where M isin its 3 state, donating 3 electrons as in the formalism M3 @C823-). The vertical scale is arbitrary in energy and in each diagram it is not to scale with respect to the others. The electronic structures of Mm@C2n with metals donating other than three electrons to the fullerene cage are not as well understood, but the disclosed purification procedures are modifiable to accommodate these potentially differentMm@C2n species, demonstrated here by the separation of Tmm@C2n endohedral fullerenes. The disclosed invention is applicable to the separation of mixed multi-element containing endohedral fullerenes. These include the endohedralspecies recently discovered (Stevenson, 1999), for example Sc3N@C80, ErSc2N@C80, Er2ScN@C80, Er3N@C80, and such. Modifications to accommodate such species include but are not limited to different oxidizing andreducing agents, different counter-anions, different solubilizing solvents, and different solvents of recrystallization. In the disclosed procedures, the oxidative generation of the endohedral metallofullerene cations can also be performed using bulkelectrochemical techniques. The disclosed purification procedures can also be applied to fullerene species other than the endohedral metallofullerenes Mm@C2n. For example, cationic purification procedures can be applied to the separation of the higher fullerenesas a general class (C2n with n≥38) from the more abundant fullerenes C60 and C70 (in the absence or presence of endohedral fullerenes). The empty higher fullerenes as a group have oxidation potentials ranging from ca. 0.7 to 0.9V, ca. 0.5 lower than the oxidation potentials of C60 and C70. Similarly, protonation of empty higher fullerenes will be easier than the corresponding protonations of C60 or C70. The addition of electrophilic reagents to emptyfullerene mixtures is selective for the higher fullerenes because of their higher basicities as evidenced by lower ionization potentials and greater ease of one-electron oxidation. These two ways of forming higher fullerene cations (oxidation andprotonation) serve as chemical handles for their separation as a group from the more abundant C60 and C70. Once oxidized or protonated, they can be recrystallized as ionic salts and separated from the non-cationic neutral C60 and C70molecules, similar to the disclosed procedures outlined above and in Schemes 1 and 2. The produced material is then greatly enriched in only the higher fullerenes. This mixture of higher fullerenes is then amenable to further separation and isolationof individual fullerenes by standard chromatographic techniques (i.e. preparative HPLC) without interference from large amounts of C60 and C70. The absence of interfering C60 and C70 will greatly increase the efficiency of thechromatographic procedure by allowing higher starting solution concentrations of mixed higher fullerenes, larger HPLC injection volumes than normally feasible with mixtures containing predominantly C60 and C70, reduction in the number ofrepetitive recycling steps, and so on. The purification and separation procedures outlined in this invention is also amenable to the separation of exohedrally derivatized fullerenes and heterofullerenes. Heterofullerenes are skeletally substituted fullerenes where one or more of thefullerene's constituent carbon atoms is replaced by another element of general formula C.sub.(2n)-xAx, as in C59N, C59B, etc. Such separations are useful to segregate these species from mixtures containing the unmodified startingmaterials. Typical preparations of exohedral C60 derivatives, for example, convert only a portion of the starting C60 to derivatized C60 molecules leaving the product material as a difficult to separate intimate mixture of the desiredC60 derivative and unconverted C60. This feature is in general true of any organic, inorganic, organometallic and hetero derivatization of fullerene molecules C2n and Mm@C2n. Thus, a simple non-chromatographic purificationmethod based on this disclosed invention for the separation of exohedral fullerene derivatives from underivatized fullerene molecules is of broad general utility. The site of oxidation or electrophilic addition to the fullerene derivative orheterofullerene may or may not be the exohedral substituent or the heteroatom, or it may alternatively by the remainder of the fullerene itself or any combination thereof. The particular details will depend entirely on the specific derivatized fullereneor heterofullerene, and in either case will not affect the applicability of the purification procedure. It will be equally applicable to endohedral fullerenes that simultaneously are exohedrally derivatized and/or are heterofullerenes. Metal-containing endohedral fullerenes have important developing applications. To efficiently exploit these applications, larger amounts of endohedral metallofullerenes are needed than are currently available. Examples of applications that willbenefit from the disclosed purification procedures include: 1. A contrast-enhancing agent for use in clinical Magnetic Resonance Imaging (MRI). MRI operates by imaging the water protons in vivo using the principles and technology of Nuclear Magnetic Resonance (NMR). When species possessing magneticspin are introduced, those in proximity to water protons will decrease the relaxation times for those protons (T1 and T2), making them "stand out" relative to those water protons not under the paramagnetic influence. This influence can beouter-sphere or inner-sphere (via direct coordination of water molecules to paramagnetic metal ions). In general, metal ions with high magnetic moments (such as the lanthanide Gd) make the best relaxation agents. However, these ions are toxic ifreleased directly into the bloodstream. Coordination chemistry has been used to minimize free Gd ions while still allowing the complex to act as a MRI relaxation agent. Endohedral Gd fullerenes have an inherent advantage over chelated metalcoordination complexes because the endohedral metal is trapped in the interior of the fullerene; there is no equilibrium with an emptied-cage/free metal species. Thus, endohedral fullerenes are likely to be safer in vivo than conventional chelates, fromthe standpoint of the metal's toxicity. Endohedral fullerenes containing magnetic metals can be purified as in the disclosed processes, and derivatized or otherwise modified for pharmaceutical usage. Primarily, derivatizations that introduce water solubility are required and many waysto accomplish this have been developed. The compounds can then be used in vivo to image internal organs and disease sites. 2. Nuclear medicine agents. Metal chelates of radioactive elements can be used as radiotracers or therapeutic sources of high-energy radioactive decay in vivo. Just like with the above MRI contrast enhancing agents, strong chemical chelatorsare used to keep the often toxic metal from circulating as a free ion. Also of concern with radionucleides is the potential ejection of the metal from the chelate following the energetic radioactive decay. Endohedral fullerenes will perform better thanconventional chelates in both these regards, as the metal cannot dissociate and the cage structure may withstand certain types of decay events and product recoil. This will decrease the deleterious side effects of these nuclear medicine agents to thepatient. The disclosed purification methods work on radioactive-metal containing Mm@C2n produced by the following methods: arc synthesis with radioactive metal precursor(s) followed by purification, etc. neutron activation aftermanufacture/purification of endohedral species (Cagle, 1999). "ion implantation" of radioactive elements into the fullerene cage nuclear recoil implantation via activation of interstitially embedded elements in solid fullerene lattices. Applications for the endohedral metallofullerene radiopharmaceuticals include radiotracers (possibly capable of crossing the blood-brain barrier) for nuclear imaging, therapeutics (with the possibility of targeting via attached functionalities)and as components in fullerene coatings of medical prosthetic devices. Another pharmaceutical application for endohedral metallofullerenes is use as generators of singlet oxygen (Tagmatarchis, 2001), which can be used therapeutically in vivo to treat disease. EXAMPLE 1 Oxidative Separation and Purification of Gd@C82 The separation and purification of Gd@C82 is presented as a representative separation of a typical M@C82 species from the mixture of fullerene materials produced by arc-discharge of metal impregnated graphite rods. Gd2O.sub.3impregnated graphite rods are arced as the anode in carbon-arc style fullerene reactor under standard conditions (ca. 30 V DC, 160 A, 150 torr He gas). The produced fullerenes and metallofullerenes contained in the raw soot are sublimed overnight atabout 750° C. under vacuum (about 1×10-3 torr) onto a water-cooled condenser. This condenser is removed from the arc apparatus anaerobically and introduced into an inert atmosphere glove box, where the sublimed mixed fullerenes andGd-containing endohedral metallofullerenes are scraped from the condenser and collected. Typical yields of this sublimate material vary from 400 to 700 mg, derived from arcing three Gd2O.sub.3 impregnated graphite rods in the present arc reactor. This mixed empty fullerene/Gd metallofullerene sublimate is then subjected to the purification procedure developed for purifying M@C82 as outlined in Scheme 1. The entire purification procedure is performed anaerobically (absence of oxygen andwater) in an argon-filled inert atmosphere glove box. Each precipitation step of Scheme 1 inevitably includes some undesired empty fullerenes (C60, C70, and higher fullerenes) simply because they are present in such greater quantities than theendohedral fullerenes. However, each successive recrystallization step enriches the quantity of metallofullerene such that by the final step there are no empty fullerenes remaining. FIG. 2 is a typical LD-TOF MS (laser desorption time-of-flight mass spectrum) of the anaerobically collected sublimate, showing the broad mixture of empty and endohedral Gd-metallofullerenes. Sublimate (300 mg) is stirred vigorously overnight ino-dichlorobenzene (ODCB) (10 ml), and then filtered through a 45 μm filter. The insoluble material removed by this filtration contains a mixture of endohedral Gd-metallofullerene species Gd@C60, Gd@C70, higher Gd@C2n (primarily withn≠41), as well as the empty the small-bandgap fullerene C74 and small amounts of other empty fullerenes C2n (n≥30). The soluble filtrate contains Gd@C82 along with comparatively large amounts of empty fullerenes C2n(n≥30, n≠37) in ODCB. The first redox step of the procedure (Scheme 1) to separate Gd@C82 from the other empty fullerenes present in greater abundance is that shown in Equation 4. To the Gd@C82 and empty fullerenes C2n(n≥30, n≠37) in ODCBsolution is added 25 mg of the oxidant [Ag.sup. ][SbF6-]; this mixture is stirred vigorously for three hours. [Ag.sup. ][SbF6-] in ODCB is of sufficient oxidizing strength to remove an electron from Gd@C82, producing itsmonocation as the hexafluoroantimonate salt and solid Ag by-product. The predominantly abundant fullerenes C60 and C70 remain neutral molecules as their potentials for oxidation are too high for Ag.sup. to remove an electron from them. Otherempty fullerenes (e.g., C76, C78, C84, etc.) that have medium to high ##STR00001## oxidation potentials also remain uncharged molecules. The soluble mixture of empty fullerenes and the [Gd@C82.sup. ][SbF6-] salt are then filtered to remove the Ag metal. ##STR00002## The solubles are next recrystallized by addition of an excess volume of a nonpolar solvent such as hexanes. The ODCB solution is added dropwise to rapidly stirring to alkane solvent (100 mL total volume of hexanes, cyclohexane, or a mixture ofcyclohexane containing 5 20% decalins). The large volume of nonpolar alkane solvent as compared to the volume of ODCB drastically decreases the solvent polarity and reduces the solubilizing ability of the initial ODCB solution such that the ionic salts[Gd@C82.sup. ][SbF6-] (and any other oxidized fullerene radical cation salts if present) rapidly precipitate from solution. Most of the neutral fullerenes remain in solution in this mixed ODCB/hexanes solvent. Small amounts of the emptyfullerenes do however precipitate along with the ionic salt. This is because the overall solubility of the empty fullerenes is lower in this mixed solvent than in pure ODCB. This step concentrates the available Gd@C82 by removing a large portionof empty fullerenes. The mixture is stirred for one hour and filtered through a 45 μm filter. The filtered solid is rinsed with additional alkane solvent (10 to 50 mL). This solid contains fullerene material enriched in Gd@C82.sup. (as its SbF6- salt) combined with some empty fullerenes that are reduced in amount as compared to the starting material. The soluble filtrate which passed through thefilter contains a mixture empty fullerenes along with a trace of Gd@C82 in the ODCB/alkane solvent mixture. This precipitation process (i.e. recrystallization) can be cycled through several iterations to further enrich the content of theGd@C82 if desired (Scheme 1). In such case, the filtered solid is re-dissolved in a minimum volume of ODCB (5 to 10 mL) and re-precipitated into rapidly stirring alkane solvent (50 to 75 mL), filtered, and washed with alkane solvent as before. While this recrystallization does enrich the precipitate's Gd@C82.sup. content relative to empty fullerenes, it does slightly diminish the yield of Gd@C82.sup. in the product because a small amount of neutral Gd@C82 gets washed into thefiltrate, likely produced via one-electron reduction of Gd@C82.sup. by trace amounts of water and other nucleophilic impurities present in the ODCB and alkane solvents. Because the total percentage of Gd@C82 in the sublimate is low (ca. 1%),it is desirable to minimize manipulations that diminish the overall content of it in each step of the purification process. In the next step, the precipitated solids enriched in Gd@C82.sup. are vigorously stirred in dichloromethane solvent. Dichloromethane is a solvent with sufficient polarity to dissolve fullerene cations having appropriate weakly coordinatingcounter-anions, but it is a comparatively poor solvent for neutral fullerene molecules. Thus, stirring the precipitate as a mixture with dichloromethane extracts primarily the soluble Gd@C82.sup. along with a small quantity of empty fullerenes. In the next step (Equation 5), anion metathesis is performed to swap the weakly coordinating SbF6- anion of the [Gd@C82.sup. ][SbF6-] salt to the more strongly coordinating chloride (Cl-). To the dichloromethanesolution is added an excess quantity of tetra-n-butylammonium chloride ([n-Bu4N.sup. ][Cl-]). In CH2Cl.sub.2, the addition of soluble [n-Bu4N.sup. ][Cl-] to the [Gd@C82.sup. ][SbF6-]/empty fullerene mixture causesimmediate precipitation of the insoluble Gd@C82Cl complex. The empty fullerenes are not affected by the presence of Cl- and remain in solution, along with the metathesis product [n-Bu4N.sup. ][SbF6-] and any excess[n-Bu4N.sup. ][Cl-]. The insoluble Gd@C82Cl is removed by filtration and rinsed with CH2Cl.sub.2. ##STR00003## To recover free Gd@C82 from the Gd@C82Cl, the chloride is first removed. This is accomplished according to Equation 6. [Ag.sup. ][SbF6-] now acts as a Cl- trap. It is mixed with Gd@C82Cl in ODCB (a solvent inwhich both species are soluble), resulting in production of soluble [Gd@C82.sup. ][SbF6-] and in rapid precipitation of insoluble AgCl. The solution is filtered to remove AgCl and recrystallized with hexanes to give solid[Gd@C82.sup. ][SbF6-] of enriched purity. In the recrystallization process, solid [Gd@C82.sup. ][SbF6-] is recovered by dropwise addition of this ODCB filtrate to rapidly stirring hexanes. The insoluble[Gd@C82.sup. ][SbF6-] is removed by filtration and washed with hexanes. ##STR00004## The resulting solid salt [Gd@C82.sup. ][SbF6-] is essentially free of empty fullerene contaminants as shown by the LD-TOF MS in FIG. 3. The material analyzed in FIG. 3 is from a preliminary test of the purification procedure andas such contains small quantities of non-Gd@C82 impurities. Subsequent application of the purification method of Scheme 1 produces Gd@C82 material of higher purity. FIG. 4 shows the electronic absorption spectrum of an ODCB solutioncontaining Gd@C82 , displaying its characteristic NIR .lamda.max at ca. 1280 nm (Akasaka, 2000(a)). FIG. 5 is a differential pulse voltammogram (DPV) of purified Gd@C82.sup. in ODCB. The 1/0 redox couple is observed at 0.52 V (vs. Ag/AgNO3 in this system) and the 0/-1 redox couple is seen at 0.03 V, corresponding to an "electrochemical gap" of 0.5 V, in agreement with the literature value (Suzuki 1996). Also visible in between these redox peaks are the corresponding redoxevents of a likely minor structural isomer of Gd@C82 (in analogy to Pr@C82, Akasaka, 2000(b)). The [Gd@C82.sup. ][SbF6-] can then be used in further chemical study as is or reduced by one electron to the neutral species Gd@C82 if desired. There many chemical methods by which this can be achieved. In any such reductionstoichiometry is controlled and a relatively weak reducing agent is utilized to minimize the chance of overreduction to the anion Gd@C82-. This reduction can be performed in several ways, one of which is described by Equation 7. A reducingagent donates one electron to [Gd@C82.sup. ][SbF6-] producing neutral Gd@C82. Decamethylferrocene, Fe(CP*)2, with an E° of ca. -0.5 V (vs the ferrocene standard) is a widely available weak reducing agent (Connelly,1996) capable of performing this reduction. One useful reduction protocol is to stir a suspension of the salt [Gd@C82.sup. ][SbF6-] in acetonitrile, in which it is insoluble, with one equivalent of decamethyferrocene, (i.e.bis(pentamethyl-cyclopentadienyl)iron, Fe(CP*)2 where Cp* is the pentamethylcyclopentadienyl anion). Fe(CP*)2 has sufficient solubility in acetonitrile to effect a one-electron reduction of [Gd@C82.sup. ][SbF6-] to Gd@C82and [Fe(Cp*)2.sup. ][SbF6-]. This mixture is stirred vigorously, producing a green solution of [Fe(Cp*)2.sup. ][SbF6-] (.lamda.max=783 nm) and a suspension of insoluble Gd@C82, which is removed by filtration andrinsed with more acetonitrile, giving a final yield of 1 to 2 mg of Gd@C82. Because the starting material [Gd@C82.sup. ][SbF6-] is insoluble in acetonitrile, the conversion efficiency to Gd@C82 in the preceding procedure islikely not 100%. For most chemical manipulations of Gd@C82, this does not pose any problems. [Gd@C82.sup. ][SbF6-] Fe(CP*)2→Gd@C82 [Fe- (Cp*)2.sup. ][SbF6-] (7) When desired, an alterative procedure for the one-electron reduction of [Gd@C82.sup. ] to Gd@C82 can be performed. Solid [Gd@C82.sup. ][SbF6-] (5 mg) is dissolved in a minimum of ODCB (1 to 5 ml). One equivalent ofFe(CP*)2 is added to this ODCB solution, with stirring. One electron reduction immediately occurs, forming Gd@C82 and [Fe(Cp*)2.sup. ][SbF6-], which remain in OCDB solution. This mixture is then added dropwise to vigorouslystirring hexanes (50 mL) to precipitate the Gd@C82 and a minimum of the [Fe(Cp*)2.sup. ][SbF6-]. The filtered solid precipitate is rinsed with acetonitrile to remove remaining [Fe(Cp*)2.sup. ][SbF6-], resulting in thefinal product of solid Gd@C82. Important to note is that the recrystallization strategy of Scheme 1 was attempted in the absence of added oxidant in separate experiments. This effected partial enrichment of Gd@C82 relative to the emptyfullerenes because of its slightly lower solubility in mixed polar/nonpolar solvents. It was however not as successful as the oxidative based procedure outlined by Scheme 1. An alternate method is to suspend solid [Gd@C82.sup. ][SbF6-] in MeCN (in which it is insoluble) and stir it with one equivalent of Fe(CP*)2. Very quickly the solution turns green from the production of[Fe(Cp*)2.sup. ][SbF6-] (confirmed by UV-Visible spectroscopy, .lamda.max=783 nm). This is stirred for several hours, filtered, and the filtered solid Gd@C82 is washed with MeCN. Because the solid[Gd@C82.sup. ][SbF6-] is not soluble in MeCN, the conversion efficiency of [Gd@C82.sup. ][SbF6-] to Gd@C82 is probably not 100% in this case. This process, however, is simple and effective enough for most practical enduses of Gd@C82. This reduction can be performed under various conditions, as known to those of ordinary skill in the art, not limited to the illustrative examples herein. The procedure disclosed above for separating Gd@C82 from empty fullerenes is summarized by Scheme 1. The exact conditions used in Scheme 1 are not limiting, however. EXAMPLE 2 Oxidative Separation and Purification of "Easily Oxidizable" MmC.sub.2n Including Gdm@C2n Scheme 2 illustrates a modification to the procedure outlined in Scheme 1. Scheme 2 outlines the procedure for extracting and purifying a larger class of endohedral metallofullerenes, Mm@C2n, as opposed to the extraction of only theM@C82 species in Scheme 1. This modification is the incorporation of the [Ag.sup. ][SbF6-] oxidant during the OCDB extraction of the mixed fullerene/endohedral fullerene sublimate. This enables in situ oxidation and extraction intosolution of otherwise insoluble endohedral species M@C2n and multiple-metal containing Mm@C2n species. For example, the procedure of Scheme 1 can be modified to produce a final product containing not only Gd@C82, but a variety of Gd endohedral metallofullerenes Gdm@C2n during the initial step of ODCB extraction of the sublimate(Scheme 2). The presence of the oxidant at this stage effects the extraction of the otherwise ODCB insoluble Gdm@C2n species (converted to cations by the oxidant) along with Gd@C82 which is soluble both as a neutral and as a cationicspecies in ODCB (Equation 8). The purification procedure of the Gdm@C2n.sup. material then proceeds in a manner similar to that used for Gd@C82.sup. (Scheme 1), as outlined in Scheme 2. ##STR00005## The end solid product is a mixture of endohedral Gd-metallofullerenes Gdm@C2n, n≥36, containing very small amounts of empty fullerenes. The procedure illustrated in Scheme 2 was performed on fullerene materials produced byarc-discharge of metal impregnated graphite rods as described in Example 1. FIG. 6 shows the LD-TOF MS of the purified Gdm@C2n material. FIG. 7 is the DPV of Gdm@C2n in ODCB, displaying the rich number of redox events in this mixedGd endohedral material. ##STR00006## The separation procedures of Schemes 1 and 2 were also performed on endohedral metallofullerenes containing metals other than Gd. Examples include Ym@C2n, Lam@C2n and Tm@C2n endohedral metallofullerenes. Y is anexample of a metal that forms species formally denoted as M3 @C2n3-, like Gd does. FIG. 8 shows the resulting Ym@C2n material obtained from this sublimate after applying the Scheme 2 process to it. FIG. 9 shows a partial DPV ofthis Ym@C2n purified material in ODCB (in agreement with the literature for Y@C82 isomers, Suzuki, 1996; Akasaka, 2000(b)). Many of these non-M@C82 Mm@C2n species are highly insoluble as neutrals and thus this process increases the total yield of extractable metallo-endohedral fullerenes. This is particularly valuable since the yield ofmetallofullerenes is such a small portion of the total fullerene species production. The ensuing purification of the mixed Mm@C2n species (Scheme 2) proceeds with the same protocol as used for M@C82 (Scheme 1). Scheme 2 can also be modified as follows, to decrease the number of steps but also lowering the extraction efficiency of Mm@C2n. Instead of performing the ODCB/Ag.sup. extraction/recrystallization steps, the process may begin with adirect CH2Cl.sub.2/Ag.sup. extraction of the sublimate. This solubilizes Mm@C2n species as cations without the prior steps, but the total extracted endohedral yield is lowered due to the trapping of endohedral species in the sublimatematrix of empty fullerenes, which have low solubility in CH2Cl.sub.2. In separate experiments, the above two disclosed procedures were conducted on Gd-fullerene soots and sublimates entirely in air, not under the anaerobic conditions used in Schemes 1 and 2. In these experiments, it was found that while thepurification processes do achieve the desired separations, their efficiencies are greatly reduced. This is likely because of the air (oxygen and/or water) sensitivity of the oxidized and neutral Mm@C2n species. This air sensitivity iscomplicated by the overall low content of endohedral fullerenes in the products of the carbon arc. Given a new production method, with a higher endohedral production efficiency than the carbon arc delivers, air exposed versions of Schemes 1 and 2 may befeasible. A further modification to the disclosed protocols of Schemes 1 and 2 (which exploit the formation of the mono-cations M@C82.sup. and Mm@C2n.sup. ) is the formation of higher charged cationic species, e.g. M@C82x andMm@C2nx , where x=2, 3, etc. Suzuki et al. (Suzuki,1996) and Akasaka et al. (Akasaka, 1996) have presented electrochemical evidence for the transient formation of M@C822 dication species at potentials near 1.1 V vs. ferrocene. Chemical formation of higher charged cations would require stronger oxidants than those of Schemes 1 and 2, as well as very weakly nucleophilic anions and solvents. Such conditions in fact are similar to those used to form discrete salts ofC60.sup. and C70.sup. (Bolskar, 1997; Reed, 2000(b)). A major advantage of dications is their low solubility due to their high ionic charge. In this modification, oxidizing conditions are used which produce a mixture of endohedralMm@C2n2 species together with empty C2n and C2n.sup. species; the endohedral dications are then be selectively precipitated and recrystallized to effect their separation from the empty fullerenes analogous to the protocols ofSchemes 1 and 2. EXAMPLE 3 M@C60, M@C70, and M@C74 There is significant metallofullerene containing material from the sublimate remaining after the processes of Scheme 1 (M@C82 separation) and Scheme 2 (easily oxidizable (with trivalent-M) Mm@C2n separation) are conducted on it. Athird class of endohedral metallofullerene separable from the sublimate are the M@C2n species such as M@C60, M@C70, and M@C74, which are in general insoluble in solvents and not easily oxidizable to soluble cations. In the case oftrivalent-M containing metallofullerenes, for example as with Gd metallofullerenes, this third class of material is like that shown in FIG. 10. The major constituent is Gd@C60. Also present by nature of having similar properties as these insolubleGdm@C2n endohedrals are empty insoluble small bandgap fullerenes like C74(D3h) and C80(Ih) etc. The amounts of these empty C2n fullerenes can be diminished by oxidation with mild oxidants like those of Scheme 2. Aluminum trichloride is useful as an oxidant for this in non-reactive solvents like ODCB. FIG. 11 shows a mass spectrum ofthe Gd@C60 class prior to the oxidative depletion of insoluble empty C2n fullerenes like C74. Depletion of the empty small-bandgap fullerene C74 and other Gd@C2n (2n≥72) from this material is accomplished by treatmentwith excess AlCl3 in ODCB, producing a dark brown solution. The insolubles were filtered off, rinsed with ODCB and CH2Cl.sub.2 (to remove excess AlCl3), followed by hexane rinsing, and dried under vacuum. The resulting material as shownin FIG. 12 (similar to the material shown in FIG. 10) is now depleted of ca. 50% of the C74 (by estimating mass spectral peak area in comparison to FIG. 11). Note the very low level of C60, C70, etc., small amounts of which may bephysically trapped in the solid matrix of insoluble Gd@C60 and thus are not accessible for extraction. The insoluble class of trivalent-lanthanide containing Mm@C2n, as shown here with the example of the Gd@C60 class, is purified byremoval of all the other material that can be solubilized either as neutral molecules or as cationic molecules. Scheme 3 presents an overall outline for the general separation procedure for all the three classes of endohedral metallofullerenes startingfrom mixed sublimate. Several lanthanide M@C2n (M=Sm, Eu, Tm, Yb and possibly Er) are reported to contain metals donating two electrons to the fullerene cage (i.e. as with the formalism M2 @C2n2-). This class of endohedral metallofullerene hasproperties differing from corresponding trivalent-metal M@C2n species, including deviations from the properties of the oxidation-resistant, insoluble M@C60 class as described above. The lanthanide endohedrals of the Mm@C60 classcontaining divalent metals can be oxidized with the mild oxidants resulting in unprecedented, soluble M@C60.sup. cations. Tm@C60 is presented here as an example of this type of metallofullerene. FIG. 13 shows the LD-TOF MS of anaerobicallyproduced Tm@C2n-containing fullerene sublimate (also containing empty fullerenes) produced by the standard arc method. This sublimate material is subjected to an oxidative extraction process similar to that shown in Scheme 2. In this first example, the process is started with reaction of the sublimate in dichloromethane with AgPF6 without prior ODCB orother solvent washing. FIG. 14 shows the Tmm@C2n.sup. containing material obtained from this process analogous to Scheme 2. This material includes significant amounts of Tm@C60 and Tm@C70 along with other Tmm@C2n speciessoluble as cations. The Tmm@C2n endohedral fullerenes are enriched relative to empty C2n as compared to the source sublimate. These results demonstrate that the disclosed purification processes apply to endohedral fullerenes withfundamentally different electronic structures than those of the M3 @C2n3- variety. More preferably, the Tm@C2n containing arc-sublimate material in FIG. 13 is first pre-extracted by stirring in ODCB for several days. This removes most soluble C2n and Tm@C2n, etc., after which the insolubles are filtered off andrinsed with hexane. The resulting solid material is then treated with the excess oxidant AgPF6 in ODCB for 12 hours. This produces a dark brown solution containing soluble cationic Tm@C2n species, including Tm@C60.sup. andTm@C70.sup. . The dark brown solution is filtered with a 0.45 μm PTFE filter and added dropwise to stirring hexanes, inducing immediate precipitation of the solid product. This precipitate containing Tm@C60.sup. and Tm@C70.sup. iscollected by filtration, rinsed with hexanes and dried under reduced pressure producing the material shown in FIG. 15. We note that the oxidation dissolves only a portion of the ODCB-insoluble Tm@C2n-; possibly there are different Tm@C2nstructures with higher oxidation potentials and/or the intramolecular polymerization of these materials is simply more difficult to surmount. Optionally, the arc-produced soot can be used in the above process directly, without prior sublimation. Soluble Tm@C2n cationic species can be extracted from ODCB-washed anaerobically handled soot by oxidation with AgPF6, etc. andrecrystallized as described above. The natures of the metallofullerene oxidations of the easily oxidizable Tm@C2n class are of particular interest. Among trivalent metal containing M@C2n, the first oxidations result from the removal of the unpaired cage electron, asfirst proposed for M@C82, with the different cage sizes (and structures) controlling the various oxidation potentials. The first oxidation of the divalent metal containing M@C2n may have different origins. One explanation is that the firstoxidation of these M@C2n is metal-centered, with conversion of M2 to M3 independent of the fullerene cage. There may be a continuum of effects in this regard across the series of endohedral lanthanides and other endohedral elementsinside of fullerene cages. ##STR00007## REFERENCES Akasaka, T.; Wakahara, T.; Nagase, S.; Kobayashi, K.; Waelchli, M.; Yamamoto, K.; Kondo, M.; Shirakura, S.; Okubo, S.; Maeda, Y.; Kato, T.; Kako, M.; Nakadaira, Y.; Nagahata, R.; Gao, X.; Van Caemelbecke, E.; Kadish, K. M. (2000)(a). "La@C82 Anion. An Unusually Stable Metallofullerene," J. Am. Chem. Soc., 122, 9316 9317. Akasaka, T.; Okubo, S.; Kondo, M.; Maeda, Y.; Wakahara, T.; Kato, T.; Suzuki, T.; Yamamoto, K.; Kobayashi, K.; Nagase, S. (2000)(b). "Isolation andcharacterization of two Pr@C82 isomers", Chem. Phys. Leff. 319, 153 156. Akasaka, T.; Wakahara, T.; Nagase, S.; Kobayashi, K.; Waelchli, M.; Yamamoto, K.; Kondo, M.; Shirakura, S.; Maeda, Y.; Kato, T.; Kako, M.; Nakadaira, Y.; Gao, X.; VanCaemelbecke, E.; Kadish, K. M. (2001). "Structural Determination of the La@C82 Isomer," J. Phys. Chem. B., 105, 2971 2974. Anderson, M. R.; Dorn, H. C.; Stevenson, S.; Burbank, P. M.; Gibson, J. R. (1997). "The Voltammetry of Sc3@C82",J. Am. Chem. Soc. 119, 437 438. Bethune, D. S.; Johnson, R. D.; Salem, J. R.; de Vries, M. S.; Yannoni, C. S. (1993). "Atoms in carbon cages: the structure and properties of endohedral fullerenes", Nature 366, 123 128. Bolskar, R. D.; Mathur, R. S.;Reed, C. A. (1996). "Synthesis and Isolation of a Fullerene Carbocation (C76.sup. )", J. Am. Chem. Soc. 118, 13093 13094. Bolskar, R. D. (1997). "The Synthesis and Characterization of Fullerene Anion and Cation Salts", Ph.D. Dissertation,University of Southern California, Los Angeles, Calif. Boudon, C.; Gisselbrecht, J.-P.; Gross, M.; Herrmann, A.; Rutimann, M.; Crassous, J.; Cardullo, F.; Echegoyen, L.; Diederich, F. (1998). "Redox Characteristics of Covalent Derivatives of the HigherFullerenes C70, C76, and C78", J. Am. Chem. Soc. 120, 7860 7868. Cagle, D. W.; Kennel, S. J.; Mirzadeh, S.; Alford, J. M.; Wilson, L. J. (1999). "In vivo studies of fullerene-based materials using endohedral metallofullereneradiotracers", Proc. Natl. Acad. Sci. USA 96, 5182 5187. Connelly, N. G.; Geiger, W. E. (1996). "Chemical Redox Agents for Organometallic Chemistry," Chem. Rev., 96, 877 910. Diener, M. D.; Smith, C. A.; Veirs, K. D. (1997). "AnaerobicPreparation and Solvent-Free Separation of Uranium Endohedral Metallofullerenes", Chem. Mater. 9, 1773 1777. Diener, M. D.; Alford, J. M. (1998). "Isolation and properties of small-bandgap fullerenes" Nature 393, 668 671. Kikuchi, K.; Suzuki, S.;Nakao, Y.; Nakahara, N.; Wakabayashi, T.; Shiromaru, H.; Satio, K.; Ikemoto, I.; Achiba, Y. (1993). "Isolation and characterization of the metallofullerene LaC82", Chem. Phys. Lett. 216, 67 71. Kitagawa, T.; Sakamoto, H.; Takeuchi, K. (1999). "Electrophilic Addition of Polychloroalkanes to C60: Direct Observation of Alkylfullerenyl Cation Intermediates", J. Am. Chem. Soc. 121, 4298 4299. Kratschmer, W.; Lamb, L. D.; Fostiropoulos, K.; Huffman, D. R. (1990). "Solid C60: a newform of carbon", Nature 347, 354 358. Liu, S.; Sun, S. (2000). "Recent progress in the studies of endohedral metallofullerenes", J. Organomet. Chem. 599, 74 86. Nagase, S.; Kobayashi, K.; Akasaka, T. (1996). "Endohedral Metallofullerenes: NewSpherical Cage Molecules with Interesting Properties", Bull. Chem. Soc. Jpn. 69, 2131 2142. Okazaki, T.; Lian, Y.; Gu, Z.; Suenaga, K.; Shinohara, H. (2000). "Isolation and spectroscopic characterization of Sm-containing metallofullerenes", Chem.Phys. Lett. 320, 435 440. Reed, C. A. (1998). "Carboranes: A New Class of Weakly Coordinating Anions for Strong Electrophiles, Oxidants, and Superacids", Acc. Chem. Res. 31, 133 139. Reed, C. A.; Bolskar, R. D. (2000)(a). "Discrete FullerideAnions and Fullerenium Cations", Chem. Rev. 100, 1075 1120. Reed, C. A.; Kim, K.-C.; Bolskar, R. D.; Mueller, L. J. (2000)(b). "Taming Superacids: Stabilization of the Fullerene Cations HC60.sup. and C60.sup. ", Science, 289, 101 104. Ruoff, R. S.; Tse, D. S.; Malhotra, R.; Lorents, D. C. (1993). "Solubility of C60 in a Variety of Solvents", J. Phys. Chem. 97, 3379 3383. Selegue, J. P.; Shaw, J. P.; Guarr, T. F.; Meier, M. S. (1994) in Recent Advances in the Chemistry andPhysics of Fullerenes and Related Materials; Kadish, K. M.; Ruoff, R. S., Eds.; The Electrochemical Society Inc., Pennington, N.J., USA, 1274 1291. Shinohara, H.; Yamaguchi, H.; Hayashi, N.; Sato, H.; Ohkohchi, M.; Ando, Y.; Saito, Y. (1993). "Isolation and Spectroscopic Properties of Sc2@C74, Sc2@C82, and Sc2@C84", J. Phys. Chem. 97, 4259 4261. Shinohara (2000)) "Endohedral metallofullerenes," Rep. Prog. Phys. 63, 843 892. Stevenson, S.; Rice, G.; Glass, T.;Harich, K.; Cromer, F.; Jordan, M. R.; Craft, J.; Hadju, E.; Bible, R.; Olmstead, M. M.; Maitra, K.; Fisher, A. J.; Balch, A. L.; Dorn, H. C. (1999). "Small-bandgap endohedral metallofullerenes in high yield and purity", Nature 401, 55 57. Strauss, S.H. (1993). "The Search for Larger and More Weakly Coordinating Anions", Chem. Rev. 93, 927 942. Sun, D.; Huang, H.; Yang, S.; Liu, Z.; Liu, S. (1999). "A Simple Method for the Selective Enrichment of Endohedral Metallofullerenes", Chem. Mater. 11,374 377. Suzuki, T.; Maruyama, Y.; Kato, T.; Kikuchi, K.; Nakao, Y.; Achiba, Y.; Kobayashi, K.; Nagase, S. (1995). "Electrochemistry and Ab Initio Study of the Dimetallofullerene La2@C80", Angew. Chem. Int. Ed. Engl. 34, 1094 1096. Suzuki, T.; Kikuchi, K.; Oguri, F.; Nakao, Y.; Suzuki, S.; Achiba, Y.; Yamamoto, K.; Funasaka, H.; Takahashi, T. (1996). "Electrochemical Properties of Fullernolanthanides", Tetrahedron 52, 4973 4982. Tso, T. S. C. ; Wan, T. S. M.; Zhang, H.-W.; Kwong,K.-P.; Wong, T.; Shinohara, H.; Inakuma, M. (1996). "Solid Phase Extraction as a Simple Method for the Enrichment of Endohedral Metallofullerenes", Tetrahedron Lett. 37, 9249 9252. Tumanskii, B. L.; Bashilov, V. V.; Solodovnikov, S. P.; Sokolov, V.I.; Bubnov, V. P.; Laukhina, E. E.; Estrin, Y. I.; Kol'tover, V. K.; Yagubskii, E. B. (1998). "EPR Study of Reactivity of Endohedral Metallofullerenes La@C82 and Y@C82", Full. Sci. and Tech. 6, 445 451. Xiao, J.; Savina, M. R.; Martin, G.B.; Francis, A. H.; Meyerhoff, M. E. (1994). "Efficient HPLC Purification of Endohedral Metallofullerenes on a Porphyrin-Silica Stationary Phase", J. Am. Chem. Soc. 116, 9341 9342. Xie, Q.; Perez-Cordero, E.; Echegoyen, L. (1992). "ElectrochemicalDetection of C606- and C706-: Enhanced Stability of Fullerides in Solution", J. Am. Chem. Soc. 114, 3978 3980. Xie, Q.; Arias, F.; Echegoyen, L. (1993). "Electrochemically-Reversible, Single-Electron Oxidation of C60 andC70", J. Am. Chem. Soc. 115, 9818 9819. Yamamoto, K.; Funasaka, H.; Takahashi, T.; Akasaka, T. (1994). "Isolation of an ESR-Active Metallofullerene of La@C82", J. Phys. Chem. 98, 2008 2011. Yang, Y.; Arias, F.; Echegoyen, L.; Chibante, L.P. F.; Flanagan, S.; Robertson, A.; Wilson, L. J. (1995). "Reversible Fullerene Electrochemistry: Correlation with the HOMO-LUMO Energy Difference for C60, C70, C76, C78, and C84", J. Am. Chem. Soc. 117, 7801 7804. Yeretzian,C.; Wiley, J. B.; Holczer, K.; Su, T.; Nguyen, S.; Kaner, R. B.; Whetten, R. L. (1993). "Partial Separation of Fullerenes by Gradient Sublimation", J. Phys. Chem. 97, 10097 10101. All references cited herein are incorporated by reference herein to the extent that they are not inconsistent with the disclosures herein and as if each individual reference had been specifically incorporated by reference. Those of ordinary skill in the art will appreciate that reagents, starting materials, solvents, reaction conditions (e.g., temperatures, pressures, etc.), purification methods, and analytical techniques and other methods and procedures other thanthose specifically disclosed herein can be applied to the practice of this invention without resort to undue experimentation. All such methods, reagents, starting materials and conditions are intended to be encompassed within the scope of thisinvention. Those of ordinary skill in the art will be aware of a number of methods, reagents, materials, procedures and techniques that are functionally equivalent to the methods, reagents, materials, procedures and techniques that are disclosed and/ordescribed herein. All such functional equivalents are intended to be encompassed by the scope of this invention. Applicants note that the instant invention is particularly useful for purifying fullerenes for derivatization and solubilization as disclosed in concurrently filed U.S. patent application entitled "Derivatization and Solubilization of InsolubleClasses of Fullerenes," application Ser. No. 10/263,375 and provisional patent applications Nos. 60/326,353 and 60/371,380. * * * * * Other References
|
| ||||||||||||||
